Graph machine learning 工具

OGB: Open Graph Benchmark（斯坦福开源用于网络神经百万量级OGB基准测试的数据集）

https://ogb.stanford.edu/

OGB is a collection of benchmark datasets, data-loaders and evaluators for graph machine learning in PyTorch.

Data-loaders are fully compatible with PyTorch Geometric (PYG) and Deep Graph Library (DGL). The goal is to have an easily-accessible standardized large-scale benchmark datasets to drive research in graph machine learning.

Deep Graph Library (DGL)

https://www.dgl.ai/

https://github.com/dmlc/dgl

DGL works on PyTorch 0.4.1+ and MXNet nightly build

PyTorch Geometric (PYG)

https://pytorch-geometric.readthedocs.io/en/latest/

https://github.com/rusty1s/pytorch_geometric

以节点预测为例，OGB支持PYG图和DGL图来表示学习框架中的数据加载方法。加载代码如下：
PYG
from ogb.nodeproppred.dataset_pyg import PygNodePropPredDataset

dataset = PygNodePropPredDataset(name = d_name) 
num_tasks = dataset.num_tasks # obtaining number of prediction tasks in a dataset

splitted_idx = dataset.get_idx_split()
train_idx, valid_idx, test_idx = splitted_idx["train"], splitted_idx["valid"], splitted_idx["test"]
graph = dataset[0] # pyg graph object

DGL
from ogb.nodeproppred.dataset_dgl import DglNodePropPredDataset

dataset = DglNodePropPredDataset(name = d_name)
num_tasks = dataset.num_tasks # obtaining number of prediction tasks in a dataset

splitted_idx = dataset.get_idx_split()
train_idx, valid_idx, test_idx = splitted_idx["train"], splitted_idx["valid"], splitted_idx["test"]
graph, label = dataset[0] # graph: dgl graph object, label: torch tensor of shape (num_nodes, num_tasks)

同时，该项目提供了一些示例代码来评估每个数据集。具体如下：
from ogb.nodeproppred import Evaluator

evaluator = Evaluator(name = d_name) 
print(evaluator.expected_input_format) 
print(evaluator.expected_output_format)

用户可以了解此数据集的输入和输出的特定格式。然后，用户可以将输入字典传递给计算器，以查看实际性能：
# In most cases, input_dict is
# input_dict = {"y_true": y_true, "y_pred": y_pred} 
result_dict = evaluator.eval(input_dict)
据悉，OGB已经正式指定上海AWS人工智能研究院的主要开源框架DGL作为数据导入平台之一。目前，DGL与PyTorch和MxNet作为后端引擎兼容，TensorFlow也在开发中。事实上，DGL已经做了很长一段时间的异构图形和可伸缩性工作，因此下一步可能是在相关领域将新技术与OGB结合起来，促进开源框架的发展。

PyGSP：Graph Signal Processing in Python

https://pygsp.readthedocs.io/en/stable/index.html

https://pygsp.readthedocs.io/en/stable/reference/index.html

Development: https://github.com/epfl-lts2/pygsp.git

https://github.com/wangg12/pygsp.git

networkx

https://pypi.org/project/networkx/

https://github.com/networkx/networkx

Website : http://networkx.github.io/

igraph：network analysis tools. igraph can be programmed in R, Python, Mathematica and C/C++.

https://igraph.org/

graph-tools，Efficient network analysis